Three Keys to the Radiation of Angiosperms Into Freezing Environments

Home , Abrophyllum, Acanthocalyx, Acicarpha, Alluaudia, Anisophyllea, Anisoptera (plant)

LETTER doi:10.1038/nature12872

Three keys to the radiation of angiosperms into freezing environments

Amy E. Zanne1,2, David C. Tank3,4, William K. Cornwell5,6, Jonathan M. Eastman3,4, Stephen A. Smith7, Richard G. FitzJohn8,9, Daniel J. McGlinn10, Brian C. O’Meara11, Angela T. Moles6, Peter B. Reich12,13, Dana L. Royer14, Douglas E. Soltis15,16,17, Peter F. Stevens18, Mark Westoby9, Ian J. Wright9, Lonnie Aarssen19, Robert I. Bertin20, Andre Calaminus15, Rafae¨l Govaerts21, Frank Hemmings6, Michelle R. Leishman9, Jacek Oleksyn12,22, Pamela S. Soltis16,17, Nathan G. Swenson23, Laura Warman6,24 & Jeremy M. Beaulieu25

Early flowering plants are thought to have been woody species to greater heights: as path lengths increase so too does resistance5. restricted to warm habitats1–3. This lineage has since radiated into Among extant strategies, the most efficient method of water delivery almost every climate, with manifold growth forms4. As angiosperms is through large-diameter water-conducting conduits (that is, vessels spread and climate changed, they evolved mechanisms to cope with and tracheids) within xylem5. episodic freezing. To explore the evolution of traits underpinning Early in angiosperm evolution they probably evolved larger conduits the ability to persist in freezing conditions, we assembled a large for water transport, especially compared with their gymnosperm cousins14. species-level database of growth habit (woody or herbaceous; 49,064 Although efficient in delivering water, these larger cells would have species), as well as leaf phenology (evergreen or deciduous), diameter impeded angiosperm colonization of regions characterized by episodic of hydraulic conduits (that is, xylem vessels and tracheids) and climate freezing14,15, as the propensity for freezing-induced embolisms (air bub- occupancies (exposure to freezing). To model the evolution of spe- bles produced during freeze/thaw events that block hydraulic pathways) cies’ traits and climate occupancies, we combined these data with an increases as conduit diameter increases5. Three evolutionary solutions unparalleled dated molecular phylogeny (32,223 species) for land liidae plants. Here we show that woody clades successfully movedintofreezing- Magno prone environments by either possessing transport networks of small 5 M o safe conduits and/or shutting down hydraulic function by dropping n o c leaves during freezing. Herbaceous species largely avoided freezing o ty le periods by senescing cheaply constructed aboveground tissue. Growth d e o 6 a n id e habit has long been considered labile , but we find that growth habit s a o e r r was less labile than climate occupancy. Additionally, freezing envir- e p u onments were largely filled by lineages that had already become herbs S or, when remaining woody, already had small conduits (that is, the trait evolvedbefore the climate occupancy). By contrast, most deciduous woody lineages had an evolutionary shift to seasonally shedding their leaves only after exposure to freezing (that is, the climate occupancy evolved before the trait). For angiosperms to inhabit novel cold environments they had to gain new structural and functional trait solutions; our results suggest that many of these solutions were

probably acquired before their foray into the cold.

Flowering plants (angiosperms) today grow in a vast range of envir- S

onmental conditions, with this breadth probably related to their diverse a

7 r i d morphology and physiology . However, early angiosperms are gen- a erally thought to have been woody and restricted to warm understory e habitats1–3. Debate continues about these assertions, in part because of Figure 1 | Time-calibrated maximum-likelihood estimate of the molecular the paucity of fossils and uncertainty in reconstructing habits for these 8–11 phylogeny for 31,749 species of seed plants. The four major angiosperm first representatives . Nevertheless, greater mechanical strength of lineages discussed in the text are highlighted: Monocotyledoneae (green), woody tissue would have made extended lifespans possible at a height Magnoliidae (blue), Superrosidae (brown) and Superasteridae (yellow). necessary to compete for light12,13. A major challenge resulting from Non-seed plant outgroups (that is, bryophytes, lycophytes and monilophytes) increased stature is that hydraulic systems must deliver water at tension were removed for the purposes of visualization.

1Department of Biological Sciences, George Washington University, Washington DC 20052, USA. 2Center for Conservation and Sustainable Development, Missouri Botanical Garden, St Louis, Missouri 63121, USA. 3Department of Biological Sciences, University of Idaho, Moscow, Idaho 83844, USA. 4Institute for Bioinformatics and Evolutionary Studies, University of Idaho, Moscow, Idaho 83844, USA. 5Department of Ecological Sciences, Systems Ecology, de Boelelaan 1085, 1081 HV Amsterdam, the Netherlands. 6Evolution & Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, New South Wales 2052, Australia. 7Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, Michigan 48109, USA. 8Department of Zoology and Biodiversity Research Centre, University of British Columbia, Vancouver, British Columbia V6T1Z4, Canada. 9Department of Biological Sciences, Macquarie University, Sydney, New South Wales 2109, Australia. 10Department of Biology and the Ecology Center, Utah State University, Logan, Utah 84322, USA. 11Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, Tennessee 37996, USA. 12Department of Forest Resources, University of Minnesota, St Paul, Minnesota 55108, USA. 13Hawkesbury Institute for the Environment, University of Western Sydney, Penrith, New South Wales 2751, Australia. 14Department of Earth and Environmental Sciences, Wesleyan University, Middletown, Connecticut 06459, USA. 15Department of Biology, University of Florida, Gainesville, Florida 32611, USA. 16Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611, USA. 17Genetics Institute, University of Florida, Gainesville, Florida 32611, USA. 18Department of Biology, University of Missouri—St Louis, St Louis, Missouri 63121, USA. 19Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6, Canada. 20Department of Biology, College of the Holy Cross, Worcester, Massachusetts 01610, USA. 21Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3AB, United Kingdom. 22Polish Academy of Sciences, Institute of Dendrology, 62-035 Kornik, Poland. 23Department of Plant Biology and Ecology, Evolutionary Biology and Behavior, Program, Michigan State University, East Lansing, Michigan 48824, USA. 24Institute of Pacific Islands Forestry, USDA Forest Service, Hilo, Hawaii 96720, USA. 25National Institute for Mathematical & Biological Synthesis, University of Tennessee, Knoxville, Tennessee 37996, USA.

6 FEBRUARY 2014 | VOL 506 | NATURE | 89 ©2014 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER seemingly arose to address the challenges of freezing: (1) woody species .0 uC across a species’ range; and ‘freezing exposed’, encountering withstood freezing temperatures without serious loss of hydraulic func- temperatures #0uC somewhere across a species’ range. This dichotomy tion by building safe water-transport networks consisting of small-diameter assumes that climate tracking through environmental changes is more conduits; (2) woody species shut down hydraulic function by becom- common than the evolution of climate occupancy; this is more likely to ing deciduous, dropping leaves during freezing periods; and (3) herb- be true if freezing exposure has a physiological cost in regions without aceous species largely avoided freezing by senescing cheaply constructed freezing21. Species were further distinguished by leaf phenology (deciduous aboveground tissue and overwintering, probably as seeds or underground or evergreen); conduit diameter (large $0.044 mm, or small ,0.044 mm; storage organs. However, the order in which angiosperms are likely to as 0.044 mm diameter is the diameter above which freezing-induced have acquired these solutions relative to exposure to and persistence in embolisms are believed to become frequent at modest tensions22); and the cold16 remains unclear. growth form (woody or herbaceous, with woody species defined as Proportions of herbaceous species, deciduous species and those with those maintaining a prominent aboveground stem that is persistent small water-conducting conduits increase towards the poles1,4,17,18, and over time and with changing environmental conditions; see Extended an earlier limited survey of angiosperm families indicated that herba- Data Fig. 1 for examples of angiosperms with woody growth habits as ceousness and ability to cope with freezing evolved in parallel19.However, we define them, and Extended Data Table 1 for a breakdown of growth exactly how global-scale ecological patterns are linked to functional evolu- habit by order within angiosperms). tion of angiosperms is uncertain. We dissect the contributions of different Among woody species we asked whether evolutionary transitions evolutionary solutions allowing angiosperms to cope with periodic freez- between climate occupancy states were significantly associated with shifts ing and assess likely pathways by which clades acquired these traits (that is, in leaf phenology and/or conduit diameter. Among all angiosperms we timing of evolution in climate occupancy relative to trait evolution). asked whether evolutionary transitions between climate occupancy states We compiled a very large species-level database of angiosperm growth were significantly associated with shifts in growth form. We determined habits (49,064 species, which is 16.4% of accepted land plant species20 the relative lability of climate occupancy (exposure to freezing) versus in The Plant List; http://www.theplantlist.org), leaf phenology, conduit traits (growth form, leaf phenology or conduit diameter) by summing diameter and freezing climate exposure. To trace species trait and climate all climate occupancy transitions and dividing by the sum of all trait occupancy relationships over evolutionary time, we generated an unpar- transitions. We also devised a novel summary based on these evolutio- alleled time-scaled molecular phylogeny for 32,223 land plant species nary transition rates that provides the likeliest pathways from the pur- in our database (Fig. 1; http://www.onezoom.org/vascularplants_tank ported early angiosperm (woody, evergreen, with large conduits and 2013nature.htm). This timetree gives us the most comprehensive view freezing unexposed) to a plant with traits for freezing conditions. yet into the evolutionary history of angiosperms. On the basis of their Because evolutionary rates are unlikely to be uniform at this phylogenetic geographic distributions, we classified species’ climate occupancies with scale, we ran growth form analyses both across the entire angiosperm respect to freezing: ‘freezing unexposed’, only encountering temperatures data set and also within each of four major lineages: Monocotyledoneae

a b N = 2,630 N = 860

Freezing Freezing Transitions per 16% 18% 0% 43% million years <0.01

Evergreen Deciduous Large conduits Small conduits 0.01–1.00

59% 7% 2% 55% >1.00

Not freezing Not freezing

c d Evergreen Large conduits Not Freezing Not freezing Persistence time (million years) Deciduous <20 Not Freezing Evergreen Small conduits Large conduits Freezing Freezing Not freezing 20–35

Small conduits >35 Deciduous Freezing Freezing

Trait first Simultaneous Climate first Trait first Simultaneous Climate first 35.4% 14.6% 50.0% 82.7% 0.0% 17.3% Figure 2 | Coordinated evolutionary transition rates between leaf defined as the inverse of the sum of the transition rates away from a given phenology or conduit diameter and climate occupancy. a, b,A character state (that is, the inverse of the sum of all arrow rates out of a character representation of coordinated evolution for the best likelihood-based state). c, d, The relative likelihood of the different pathways out of the evergreen model between leaf phenology for 2,630 species (evergreen, dark green; and freezing-unexposed state and into the deciduous and freezing-exposed deciduous, light green) and climate occupancy (freezing exposed (freezing), state (c), and out of the large-diameter conduit and freezing-unexposed state striped; freezing unexposed (not freezing), solid) (a), and conduit diameter for and into the small-diameter conduit and freezing-exposed state (d). The three 860 species (large ($0.044 mm), light blue; small (,0.044 mm), dark blue) and possible pathways between two focal character state combinations provide climate occupancy (b) based on models fit to all Angiospermae. The sizes of the insight into whether lineages typically evolved: (1) with the trait first, such that black arrows in the plot are proportional to the transition rates between each phenology or conduit diameter shifted before encountering freezing; (2) with possible state combination (larger arrows denote higher rates; no arrows for climate occupancy first, such that phenology or conduit diameter shifted after rates of 0). The number at the top of each panel denotes the number of extant encountering freezing; or (3) both simultaneously, such that shifts in phenology Angiospermae species used in the analyses and percentages denote the or conduit diameter and encountering freezing happened at the same time (see percentage of extant species with that character state. The size of each circle is Supplementary Information for further details). proportional to the persistence time in that state, where persistence time is

(monocots), Magnoliidae (magnoliids), Superrosidae (superrosids) . monocots). Of 104 models evaluated, a 40-parameter model allow- and Superasteridae (superasterids) (see ref. 10 for lineage definitions); ing each major lineage to have its own transition matrix received most these clades represent , 22%, 3%, 34% and 34%, respectively, of all support (Extended Data Table 4). These results were generally robust extant angiosperm species. to uncertainty about whether species in the freezing-unexposed state Across woody angiosperms, a model that assumed coordinated evolu- actually lacked an ability to cope with freezing (Supplementary Informa- tion of leaf phenology and climate occupancy was strongly supported tion). Across angiosperms, asymmetry of transition rates led to numer- over a model that assumed they evolved independently (Akaike infor- ous extant species in the woody freezing-unexposed and herbaceous mation criteria (DAIC) 5 310.1; Fig. 2a and Extended Data Table 2). freezing-exposed states (Fig. 3a and Extended Data Table 3). The large Deciduous freezing-exposed and evergreen freezing-unexposed were number of extant species in the woody freezing-unexposed state, accord- highly persistent character states (Fig. 2a, as indicated by size of the ing to our model, was the result of this state being persistent (Fig. 3a). circles, and Extended Data Table 3); persistence times (that is, expected Even within monocots, where relatively few woody species exist, the time until state change) are defined as the inverse of the sum of estimated woody freezing-unexposed state was strongly persistent. The herbaceous transition rates away from a given character state. Therefore, in the freezing-exposed state, on the other hand, had low persistence times, presence of freezing, the deciduous state was far more stable than the indicating that the numerous extant species (N 5 4,066 out of 12,706 evergreen one. We also found that leaf phenology was generally about species for which data are available) were due to many rapid transitions as labile as climate occupancy (climate:trait rate ratio 5 0.845), and it both into and out of this character state (Fig. 3a). Climate occupancy was was also far more likely to evolve as a response to a change in envir- much more labile than growth form (climate:trait rate ratio 5 4.93). onment rather than arising before encountering freezing (that is, cli- Furthermore, the predominant pathway within angiosperms from the mate occupancy evolved first; Fig. 2c). woody freezing-unexposed state to the herbaceous freezing-exposed Similarly, across woody angiosperms, a model assuming coordinated state was to first evolve the herbaceous habit and subsequently enter evolution of conduit diameter size and climate occupancy was strongly habitats with freezing-exposed conditions (that is, the trait evolved supported over a model that assumed they evolved independently before the climate occupancy; Fig. 3b). This, in combination with the (DAIC 5 21.5; Fig. 2b and Extended Data Table 2). Both climate occu- conduit diameter results, suggests that lineages that successfully colo- pancy states (freezing exposed and freezing unexposed) in combina- nized new freezing environments were probably predisposed to do so, tion with small conduits were highly persistent (Fig. 2b and Extended at least for these two traits. Data Table 3). Additionally, no species with large conduits were in the Although our focus here is on evolutionary links between species freezing-exposed state, indicating that this is a highly transitory char- distributions with respect to freezing conditions and traits that allow acter state (that is, short persistence time). As with leaf phenology, species to cope with freezing, we note that differential diversification climate occupancy and conduit diameter were similar in their overall rates23 and vagility among lineages also certainly played their parts in lability (climate:trait rate ratio 5 0.895); however, a shift into environ- determining why we see species where we do today. For instance, herbs ments with freezing temperatures was far more likely to occur after may have higher speciation and/or extinction rates than woody taxa24. conduits had already shifted from large to small (that is, the trait evolved Additionally, growth form may influence a plant’s ability to disperse to before climate occupancy; Fig. 2d). and colonize newly emerging locations with freezing temperatures25. Evolutionary shifts in growth habit were also strongly coordinated Tests of these alternatives are critical for fully understanding how angios- with shifts in climate. However, the nature of coordination varied con- perms radiated into freezing environments, but such analyses require siderably among major angiosperm clades (Extended Data Table 3), as an even more complete record of global distributions of vagility and did overall transition rates (superrosids and superasterids . magnoliids growth habit across land plants and a comparably more completely

a Freezing Figure 3 | Coordinated evolutionary transition 17% 32% rates between growth form and climate occupancy. a, A representation of coordinated Angiospermae Herbaceous Woody (N = 12,706) evolution for the best likelihood-based model between growth form for 12,706 species

40% 11% (herbaceous, green; woody, brown) and climate occupancy based on a model assuming the same Not freezing rates were applied to all Angiospermae (top plot above the dashed arrow), and the best-fit model, in which rates were estimated separately for the major 1% 64% 14% 2% 24% 14% 21% 34% lineages, that is, Monocotyledoneae, Magnoliidae, Superrosidae and Superasteridae (bottom four Monocot. Magnoliidae Superrosidae Superasteridae (N = 2,873) (N = 532) (N = 4,763) (N = 4,017) plots below the dashed arrows). b, The weighted average (by clade diversity) of the relative 5% 30% 82% 3% 59% 3% 37% 9% likelihood of the different pathways out of the woody and freezing-unexposed state and into the herbaceous and freezing-exposed state (see Fig. 2 and Methods for further details). b Woody Not freezing Transitions per Persistence time million years (million years)

<0.01 <20 Herbaceous Woody Not freezing Freezing 20–60 0.01–0.10

>0.10 >60

Herbaceous Freezing

Trait first Simultaneous Climate first 58.0% 0.0% 42.0%

6FEBRUARY2014|VOL506|NATURE|91 ©2014 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER sampled phylogeny. These are non-trivial improvements as we currently 2. Wing, S. L. & Boucher, L. D. Ecological aspects of the Cretaceous flowering plant 20 radiation. Annu. Rev. Earth Planet. Sci. 26, 379–421 (1998). have growth habit data for only 16% of accepted land plants (R.G.F. 3. Feild,T.S.,Arens,N.C.,Doyle,J.A.,Dawson,T.E.&Donoghue,M.J.Darkanddisturbed: et al., manuscript submitted) and molecular and climate data for 26% a new image of early angiosperm ecology. Paleobiology 30, 82–107 (2004). (12,706 species) of those taxa. Total trait records are fewer for pheno- 4. Moles, A. T. et al. Global patterns in plant height. J. Ecol. 97, 923–932 (2009). 5. Tyree, M. T. & Zimmermann, M. H. Xylem Structure and the Ascent of Sap (Springer, logy (6,705 species) and conduit diameter (2,181 species). 2002). Among three key angiosperm strategies successful in today’s freez- 6. Cronquist, A. The Evolution and Classification of Flowering Plants. (Houghton Mifflin, ing environments (deciduous leaves, small conduits and herbaceous 1968). 7. Kattge, J. et al. TRY—a global database of plant traits. Glob. Change Biol. 17, habit), our analyses indicated two especially striking findings. First, the 2905–2935 (2011). pathway to herbaceousness or small conduits in freezing environments 8. Stebbins, G. L. The probable growth habit of the earliest flowering plants. Ann. Mo. largely involved acquisition of the trait first (followed by adaptation Bot. Gard. 52, 457–468 (1965). 9. Taylor, D. & Hickey, L. Phylogenetic evidence for the herbaceous origin of to a new climate), whereas the pathway to deciduousness in freezing angiosperms. Plant Syst. Evol. 180, 137–156 (1992). environments was largely via a shift in climate occupancy first (fol- 10. Soltis, D. E. et al. Angiosperm phylogeny: 17 genes, 640 taxa. Am. J. Bot. 98, lowed by evolution of the trait). Second, transitions between growth 704–730 (2011). 26 11. Smith, S. A., Beaulieu, J. M. & Donoghue, M. J. An uncorrelated relaxed-clock habit states should be fairly simple genetically , involving suppression analysis suggests an earlier origin for flowering plants. Proc. Natl Acad. Sci. USA 27 and re-expression of only a few genes , and, traditionally, growth habit 107, 5897–5902 (2010). has been considered highly labile (ref. 6, but see refs 16, 28, 29). Our 12. Spicer, R. & Groover, A. Evolution of development of vascular cambia and secondary growth. New Phytol. 186, 577–592 (2010). results are consistent with climate occupancy being more labile than 13. Feild, T. S. & Wilson, J. P. Evolutionary voyage of angiosperm vessel structure- growth habit, and freezing environments being largely filled by a subset function and its significance for early angiosperm success. Int. J. Plant Sci. 173, of lineages that were already herbaceous or, if woody, had small con- 596–609 (2012). duits before they encountered freezing. Why these lineages initially evolved 14. Philippe, M.et al. Woodyornot woody? Evidence forearlyangiosperm habitfromthe Early Cretaceous fossil wood record of Europe. Palaeoworld 17, 142–152 (2008). a herbaceous habit and small conduit sizes remains unclear; these traits 15. Wiens, J. J. & Donoghue, M. J. Historical biogeography, ecology and species are probably tightly associated with responses to other environmental richness. Trends Ecol. Evol. 19, 639–644 (2004). gradients (for example, aridity in the tropics) and numerous other aspects 16. Donoghue, M. J. A phylogenetic perspective on the distribution of plant diversity. Proc. Natl Acad. Sci. USA 105, 11549–11555 (2008). of a plant’s ecological strategy (for example, seed size, tissue defence, and 17. Wheeler, E. A., Baas, P. & Rodgers, S. Variations in dicot wood anatomy: a global so on) related to resource acquisition and disturbance regimes. Therefore, analysis based on the Insidewood database. IAWA J. 28, 229–258 (2007). successful shifts between stem constructions take more than just turn- 18. Botta, A., Viovy, N., Ciais, P., Friedlingstein, P. & Monfray, P. A global prognostic scheme of leaf onset using satellite data. Glob. Change Biol. 6, 709–725 (2000). ing on or off a few genes. 19. Judd, W. S., Sanders, R. W. & Donoghue, M. J. Angiosperm family pairs: preliminary By weaving together a series of disparate threads encapsulating evolu- phylogenetic analysis. Harv. Pap. Bot. 5, 1–49 (1994). tion, functional ecology and the biogeographic history of angiosperms, 20. Paton, A. J. et al. Towards target 1 of the global strategy for plant conservation: a working list of all known plant speciesprogress and prospects. Taxon 57, 602–611 including extensive functional trait databases and an exceptionally (2008). large timetree, we have documented the likely evolutionary pathways 21. Loehle, C. Height growth rate tradeoffs determine northern and southern range of trait acquisition facilitating angiosperm radiation into the cold. limits for trees. J. Biogeogr. 25, 735–742 (1998). 22. Davis, S. D., Sperry, J. S. & Hacke, U. G. The relationship between xylem conduit diameter and cavitation caused by freezing. Am. J. Bot. 86, 1367–1372 (1999). METHODS SUMMARY 23. Maddison, W. P. Confounding asymmetries in evolutionary diversification and To examine the evolutionary responses to freezing in angiosperms, we first com- character change. Evolution 60, 1743–1746 (2006). piled trait data on leaves and stems from existing databases and the literature. 24. Soltis, D. E. et al. Phylogenetic relationships and character evolution analysis of Saxifragales using a supermatrix approach. Am. J. Bot. 100, 916–929 (2013). Growth form data came from numerous sources and were coded as a binary trait 25. Thomson, F. J., Moles, A. T., Auld, T. D. & Kingsford, R. T. Seed dispersal distance is (woody or herbaceous; Supplementary Table 1). Leaf phenology and conduit dia- more strongly correlated with plant height than with seed mass. J. Ecol. 99, meter came from existing databases (see Supplementary Information for a list). 1299–1307 (2011). Second, taxonomic nomenclature was made consistent among data sets and up to 26. Groover, A. T. What genes make a tree a tree? Trends Plant Sci. 10, 210–214 (2005). date by querying species names against the International Plant Names Index 27. Lens, F., Smets, E. & Melzer, S. Stem anatomy supports Arabidopsis thaliana as a model for insular woodiness. New Phytol. 193, 12–17 (2012). (http://www.ipni.org/), Tropicos (http://www.tropicos.org/), The Plant List (http:// 28. Jansson, R., Rodrı´guez-Castan˜eda, G. & Harding, L. E. What can multiple www.theplantlist.org/) and the Angiosperm Phylogeny website (http://www.mobot. phylogenies say about the latitudinal diversity gradient? A new look at the tropical org/MOBOT/research/APweb/). Third, we obtained species’ spatial distributions conservatism, out-of-the-tropics and diversification rate hypotheses. Evolution 67, from Global Biodiversity Information Facility records (http://www.gbif.org/; Sup- 1741–1755 (2013). plementary Table 4) and then determined whether species encountered freezing 29. Beaulieu, J. M., O’Meara, B. C. & Donoghue, M. J. Identifying hidden rate changes in the evolution of a binary morphological character: the evolution of plant habit in temperatures using climate data from the WorldClim database (http://www.worldclim. campanulid angiosperms. Syst. Biol. 62, 725–737 (2013). org/). Fourth, we constructed a dated phylogeny for these species by downloading available GenBank sequences (http://www.ncbi.nlm.nih.gov/genbank/) for seven Supplementary Information is available in the online version of the paper. gene regions. Genetic data were compiled and aligned using the PHLAWD pipe- Acknowledgements We thank T. Robertson and A. Hahn at the Global Biodiversity line (v.3.3a), and maximum-likelihood-based phylogenetic analyses of the total Information Facility for providing species’ georeference points, A. Ordonez for sequence alignment were performed using RAxML (v.7.4.1), partitioned by gene providing growth form data, and A. Miller and D. Ackerly for helpful comments on a region and with major clades (that is, families and orders) constrained according to draft of this manuscript. Support for this work was given to the working group ‘‘Tempo and Mode of Plant Trait Evolution: Synthesizing Data from Extant and Extinct the APG III classification system. Branch lengths were time-scaled using congrui- Taxa’’ by the National Evolutionary Synthesis Center (NESCent), National Science fication, which involved using divergence times estimated from a reanalysis of a Foundation grant #EF- 0905606 and Macquarie University Genes to Geoscience broadly sampled data set (Extended Data Fig. 2 and Supplementary Tables 2 and 3). Research Centre. Last, tests of coordinated evolution among traits in our database were analysed in Author Contributions A.E.Z., W.K.C., D.C.T. and J.M.B. designed the initial project, wrote the corHMM R package; transition rates between two binary traits were analysed the original manuscript and carried out analyses. J.M.E., S.A.S. and D.C.T. constructed using a likelihood-based model. the timetree. J.M.E., R.G.F., D.J.M., B.C.O’M. and S.A.S. were major quantitative contributors, especially with the development of new methods, analyses, graphics and Online Content Any additional Methods, Extended Data display items and Source writing. A.T.M., P.B.R., D.L.R., D.E.S., P.F.S., I.J.W. and M.W. were large contributors Data are available in the online version of the paper; references unique to these through the development of initial ideas, methods, dataset curation, analyses and sections appear only in the online paper. writing. L.A., R.I.B., A.C., R.G., F.H., M.R.L., J.O., P.S.S., N.G.S. and L.W. contributed data sets and discussions, and read drafts. Received 3 July; accepted 5 November 2013. Author Information Data and code are deposited at the Dryad Digital Repository Published online 22 December 2013; corrected online 3 January 2014 (see full-text (http://dx.doi.org/10.5061/dryad.63q27) and TRY (http://www.try-db.org/). Reprints HTML version for details). and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the 1. Sinnott, E. W. & Bailey, I. W. The evolution of herbaceous plants and its bearing on online version of the paper. Correspondence and requests for materials should be certain problems of geology and climatology. J. Geol. 23, 289–306 (1915). addressed to A.E.Z. ([email protected]).

Extended Data Figure 1 | Examples of the definition of ‘woody’. a–d,We Arizona, USA, c, Rhopalostylis sapida (Arecaceae) and Cyathea sp. defined ‘woody’ as having a prominent aboveground stem that is persistent (Cyatheaceae), Punakaiki, South Island, New Zealand. d, Pandanus sp. over time and with changing environmental conditions. a, Liriodendron (Pandanaceae), Moreton Bay Research Station, North Stradbroke Island, tulipifera (Magnoliaceae), Joyce Kilmer Memorial Forest, Robbinsville, North Queensland, Australia (photographs by A.E.Z.). Carolina, USA. b, Carnegiea giganteana (Cactaceae), Biosphere II, Tucson,

nordnedodbahR

m ah mul t s

Lampranthus e ullyh

ai Delosperm u

A m g

lu Bougainvillea l d o

m s

s m Sarcobatu n y alco Cal Phytolacc t uryp t d S o u a n Ba m Halophytu h e in hpo Limeu s n

n C s hP n p o u a

i o r etS Mirabilis Gisekia p o t y ip or ogy Claytonia e o ma y r e n s ro r c d n M s beuia e p y the i l ll t p B k m o c o Alluaudia Rivin e n om

e xo trot t hpir si o n e ca ga l n a Portulac mr ra s n e a

Basella mi Talinum s i a i a l t iB o c a o

a ai t d m a sun n a iL l ai r Opunt a i Hydrangea Pereski u ti a S n F xe P a he Plumbago F i a s s Tamarix g a D a T Te Mentzeli A M Drosophyllu P CurtisiaCornu Drosera aG o Nepen Nyss m Berberidopsis c A e Couroupit Impatiena e Schoepfia Polemoniu t a Opilia rameris rcgra Santalum t Ximenia alony i a Heisteria a Tetracer s Hibber o s a Dilleni a M uoF a Myrothamnu Diospyros s Gunnera uxus m q Pachysandr m anil B Ternstroemia via a Didymele x Tetracentro Androsac Trochodendron eiu t s Petrophil Maesa k r a Roupala a a PrimulaClavij a Platanu ar Nelumb ra m ai Camellia Meliosm a Sabia Ranunculu Glaucidiu Sarraceni spo Actinidi Styrax a Hydrastis m ia Nandina m Caulophyllum zlo Gala e ino Menispermu a Clethr T Lardizabal h Arbutus Decaisnea c m x s Eucommia yC Sargentodox Oncot Icacina a Circaeaster h elea r a Garrya i Hypecou cs ll a Dicentr a E s SphenocleMont Eupt a Hydrole Vahliheca Ceratophyllua Saccharu Ipomoea Ze i Oryz n a a a NicotianaCuscut i Hordeum a a Stegolepi a Sparganiumiese So Atropa a 16`` Typh a Vr Luculiala a Marant nu Mus StrychnoCo GelsemiuG 17`` Strelitzia m Tradescantia aliumffe a Philydrum e s Nerium a Elaeis a Gentiana Chamaedorea HydrophyllumExacum s Agav m Yucca Borago Plocosperma Lomandr m Ehretia 18`` Asparagu s Beaucarne Jasminum Alliu Polypremum Xanthorrhoeas m PeltantheraSyringa Iri Blandfordiidiua Saintpaulia Phalaenopsix a Calceolaria Onc a Scrophularia 20`` 19`` 15`` Lilium AntirrhinumHalleria Smil Trillium n 21`` Carludovic Sesamum Croomia Martynia 14`` Dioscorea Triglochi a Catalpa Potamogeton m Pinguicul 11 Tho Byblis 9 Alism a a Acanthus Hydrocharis a m 13`` anders 10 Pleeofieldi a T hosom Verben Spathiphyllut um PaulowniaLamium ia Xan i a Lemna s PediculaPhryma 12`` Oront 22`` Acoru s Hedycaryaarpus Helwingia ris Peumuoc PhyllonomaI lex Gyr rnandia IrvingbaileyGrisollea He Gomphandr Cryptocarya ra Laurus a GonocaryumCitronella 8 Atherospermateg a a Cardiopteris a Daphnand Gomor s Polyosm Siparun Er emo Idiospermuma Anopteru Calycanthu sy a a ne Annon tia Tribele Escallonia s Canang a s Eupoma Forgesia Galbulimimaa Pen tic taphValdivi 24`` Mauloutchia rag a Myris a n CarpodetuRoussema Degeneri Abroph a Magnoli Pseudonemacladus Liriodendro yllums Cuttsi 23`` Piper a Peperomias DialypetaluLobelia Houttuynia ia Saururustoloch Ari TracheliuCyphim 7 s a Thotteaori Ca Lact mpanulm ruma Phelline Sa a Drimys ArgophylluCorokia Platyspermatio 6 Tasmannia m Takhtajania Canella Crispilob n s Alseuosm Cinnamodendron a Chloranthua Wittsteiniia Sarcandrarina Donatiaa Asc m Stylidiu Hedyosmu Fo m rster Schisandra Villarsi a Kadsura Nympho a cium Menyantheside Illi a s Trimeni Faur 5 Austrobaileya Dampieraia Nymphaea Scaevol Nuphara Goodeni a Braseni a a Boopis Cabomba Moschopsi 4 3 Trithuri s a Acicarpha Amborellhia B elwitsc arnadesi W a Gnetum Gerbera Echinop 2 Pinus Tragopogo s Podocarpus n ia Cichorium 25`` Metasequo Lactuc Cycas a Zamia Tagetes Guizotia Ginkgo Helianthus Desfontainia Columellia Berzelia Brunia Pennantia 1 Aralidium Melanophylla Torricellia 26`` Griselinia Sollya Pittosporum 400.0 350.0 300.0 250.0 200.0 150.0 100.0 50.0 0.0 Hydrocotyle Aralia Panax Cussonia 27`` a Acalypha Heder Spathiostemo Tetrapanax n lera Ricinus Scheff Lasiocroton Polyscias Dal echampia Pseudopanaxa Conceveib raplasandr Ho a Tet malanthus Delarbreas Hura Myodocarpua Euphorbi Mackinlay Pimelodendrona tysace Pla a Codiaeum Azorell Trigonostemon Croto Saniculas n Arctopu a Manihot Hevea Heteromorph s Daucu a Moultonianthu Su Angelicm regad T a s m etrorchidiu CoriandruApiu Omphale hum m Anet a Endospermua Neoscortechinia Quintini m Clutia Paracryphia Pera rnum Pogono SphenostemonVibu a Cespedesipho Sambucus Sauvagesi ra Sinadoxxa a Ado Ochna xa Luxemburgia Tetrado Touroulia Diervilla Quiin a Weigela Medusagyna ra s Caryoca HeptacodiumLonicerpo Hirtella e ica ia 28`` Lic r ester Atunani SymphorLeyc a a Triosteum Chrysobalanua Linnae a Euphronia Abeli a Dichapetalu Dipelt Tapur s a Trigoniaa KolkwitziaZabeli a Balanop m ina Durande Mor Hugonia s Cryptothladi Linu a Reinwardti Acanthocalyx m Triplostegiabiosaa 29`` Centroplacus Sc s Bhesa Dipsacus Pand a s Galeari Patrinia a a erocephalode Microdesmis Pt Bruguier a NardostachyValerian Rhi Fedialla Carallizopho Centranthus ia Paradrype a leriane s Erythroxylua ra Va Soyaux a Ctenolophon Drype tes PeridiscuPaeoni r ltingia Putranjiva A m 30`` 31 Lophopyxites m m M 39 Austrobuxusicrantheu Liquidamba Di Petass s DaphniphylluCorylopsis A ilia m Cercidiphyllu Te nd lo r Hamamelis r st ia Podocalyxtra ost i Itea gm ExbucklandiaRhodoleia 38 Lachnostylico a a Croiz cc ch a ChoristylisRibes ys Heywoodia us e Phyllanthua B ti Saxifrag 32 i a s Heuchera m Aporusasc Sedum Dicell ho Kalancho Malpighi fia rpaea 34 37 T s ca s Acridocarpuhryallis a a tr Byrsonim e s Elatine T Penthorum Leea 36 a Aphanopetalu m Bergi Haloragi Viti Populu Myriophyllum Sali a 33 Idesi a a s Geraniu Polio x Dovyalis s elianthusViviania 35 a M a F a Abatila thyrsis Prockia c Terminali ia o OenotheraLythrumroni n Scyphos u li s Clidemie Lunania a tr pt O s Casearia ia Lozani Qualea Cry La Myrtu ba a Hybanthu c tegi Leonia istem a Hyme AphloiaIxer a a Eucalyptu Viol a Rinorea G a a Passifloroupia a na Staphylea r s Paropsia s raria t s T n Strasburgeri tachyurus rse Malesu thera Carpotroche CrossosomS Ni u s r Picramni B s Calon ne Erythrospermu Schinu a Hydnocarpu ra a a Kiggelaria he Citru Achari Swietenia P coba rbi Bix m Hypericumangium Cupaniopsi ai Tr Vismi a Ailanthu lu Tapisci m Cratoxylu iadenu m Podostemum a cret a Marathrum Dipentodon Caloph Mamme Mes a Anisoptera s A s S a B Caric m Thymelaea Batis s P r Ga c m m a Clusi e ua Irvingi nohytaea Gossypiu n Helianthemu Floerkea a Klainedox n yllu m Reseda O rcinia m Vantane e Tropaeolu ia e Hu at a iacu s chthocosmu t CappariBrassic Eucryphia p i o G a C aS a m Crinodendro h m miria s E eissois c Sloane a Quillaja lbiz s e a Gua Krameri Averrho laeo o Lotu Dapania p l Polygal A Cicer a O a s Rourea rB g Arabidopsi s Connaru Glycin s A u i Hu P n unu Euonym x nus a T Pisu Celastrus ol Brexia n ah a b a E f P g a M S a a D P S a i s r ia a s r t im carpus a s m r si i Spiraea l a sun a Pr t s o l n l ol a t i eL u Stylobasiu a a i e u a p Celti n u a a Medicag Pilea a pi n n si sit c ix s lle t o m ir pelo p r e i n

i i bru g g u c t n tisc c Moru n h a al rau

Urtic r Zelkov uc pr

r i e tya o ts h a l i yt s a k s a d a Elaeagnu o o e i Rha y o f e g r d m a h r Fagus

A o s

Ceanothus ac C Da n y u c p C o a s u u e m M Begon s s u

b saC un a o g n t u J h x C n

i y Q e s

t o a s d i i

ili

a d C suo u r

o t

h o i r r m

t Anisophyllea a n o N y n

s n

Extended Data Figure 2 | Reference timetree used for congruification indicated at the nodes with green circles, and numbers correspond to fossils analyses. Results of the divergence time estimation of 639 taxa of seed plants described in Supplementary Table 2. Concentric dashed circles represent from the reanalysis of a previously described10 phylogeny. Fossil calibrations are 100-Myr intervals as indicated by the scale bar.

Extended Data Table 1 | Number of species in different growth forms by clade

Lineage Woody Herbaceous Total Proportion herbaceous Angiospermae 28650 17347 45997 0.38 Magnoliidae 2438 75 2513 0.03 Monocotyledoneae 1226 9894 11120 0.89 Superasteridae 8468 4863 13331 0.36 Superrosidae 14885 1956 16841 0.12 ANA grade+Chloranthales Amborellales 1 0 1 0.00 Austrobaileyales 48 0 48 0.00 Chloranthales 18 7 25 0.28 Nymphaeales 0 43 43 1.00 Magnoliidae Canellales 71 0 71 0.00 Laurales 1212 6 1218 0.00 Magnoliales 1053 0 1053 0.00 Piperales 102 69 171 0.40 Monocotyledoneae Acorales 0 7 7 1.00 Alismatales 3 513 516 0.99 Arecales 793 0 793 0.00 Asparagales 141 4133 4274 0.97 Commelinales 0 180 180 1.00 Dioscoreales 0 178 178 1.00 Liliales 35 459 494 0.93 Pandanales 80 17 97 0.18 Petrosaviales 0 3 3 1.00 Poales 109 4075 4184 0.97 Zingiberales 61 329 390 0.84 Basal eudicots+Gunnerales Buxales 31 0 31 0.00 Ceratophyllales 0 3 3 1.00 Gunnerales 2 14 16 0.88 Proteales 1354 3 1357 0.00 Ranunculales 134 488 622 0.78 Trochodendrales 2 0 2 0.00 Superasteridae Apiales 410 226 636 0.36 Aquifoliales 211 0 211 0.00 Asterales 548 1775 2323 0.76 Berberidopsidales 3 0 3 0.00 Bruniales 65 0 65 0.00 Caryophyllales 545 712 1257 0.57 Cornales 163 68 231 0.29 Dilleniales 71 0 71 0.00 Dipsacales 151 61 212 0.29 Ericales 2798 350 3148 0.11 Escalloniales 23 0 23 0.00 Garryales 17 0 17 0.00 Gentianales 1508 280 1788 0.16 Lamiales 1214 1035 2249 0.46 Paracryphiales 20 0 20 0.00 Santalales 242 20 262 0.08 Solanales 254 200 454 0.44 Superrosidae Brassicales 136 389 525 0.74 Celastrales 228 11 239 0.05 Crossosomatales 31 0 31 0.00 Cucurbitales 62 169 231 0.73 Fabales 2462 448 2910 0.15 Fagales 745 0 745 0.00 Geraniales 27 63 90 0.70 Huerteales 8 0 8 0.00 Malpighiales 2978 294 3272 0.09 Malvales 1195 64 1259 0.05 Myrtales 2787 79 2866 0.03 Oxalidales 396 14 410 0.03 Picramniales 16 0 16 0.00 Rosales 1465 143 1608 0.09 Sapindales 2082 7 2089 0.00 Saxifragales 190 246 436 0.56 Vitales 42 1 43 0.02 Zygophyllales 35 12 47 0.26

Number of species that are woody, number of species that are herbaceous, total number of species, and proportion of herbaceous species in major lineages and orders. Proportions in bold are lineages with .0.5 species that are herbaceous.

Extended Data Table 2 | Coordinated evolutionary model fits for leaf phenology, conduit diameter and climate occupancy

Leaf Phenology and climate occupancy

Model Number of -lnL AIC AIC wi parameters Character independent 4 -2305.4 4618.9 312.8 <0.01 Character dependent, equal rates 1 -2401.3 4804.5 498.4 <0.01 Character dependent, all rates diff 8 -2160.0 4336.0 29.9 <0.01 Character dependent, all rates diff* 12 -2141.1 4306.1 0 0.99 Conduit diameter and climate occupancy

Model Number of -lnL AIC AIC wi parameters Character independent 4 -603.65 1223.3 21.5 <0.01 Character dependent, equal rates 1 -739.8 1481.6 279.8 <0.01 Character dependent, all rates diff 8 -592.91 1201.8 0 0.98 Character dependent, all rates diff* 12 -592.91 1209.8 8.0 0.02

The likelihood-based best model in each case (shown in bold italics) was chosen based on both AIC and Akaike weights (wi). Also listed for each model are the number of parameters, negative log likelihood (2lnL), and DAIC. The asterisk indicates a model where simultaneous changes in any two binary characters were allowed to change.

Extended Data Table 3 | Coordinated evolutionary model transition rates

The estimated transition rates for the best likelihood-based evolutionary transitions model between climate occupancy and either growth habit, leaf phenology or conduit diameter evolution are included. The numbers in parentheses denote the values at the 2.5% and 97.5% quantiles of the distribution of parameter estimates obtained from the same analyses run on the 100 bootstrapped trees (see Supplementary Information). The leaf phenology model includes transitions between combinations of leaf phenology (evergreen, deciduous) and climate occupancy (freezing exposed, freezing unexposed), the conduit diameter model includes transitions between combinations of conduit diameter (large $0.044 mm, small ,0.044 mm) and climate occupancy, and the growth habit model includes transitions between combinations of growth form (herbaceous, woody) and climate occupancy. Arrows denote the direction of the transition. The growth habit model assumes separate models for the major groups within angiosperms: Monocotyledonae, Magnoliidae, Superrosidae, Superasteridae and all remaining angiosperms (the rest), including the ANA grade, Chloranthales, Ceratophyllales and basal eudicots plus Gunnerales. The leaf phenology and conduit diameter models assume a single model for all angiosperms.

Extended Data Table 4 | Coordinated evolutionary model fits for growth form and climate occupancy

Model Number of parameters -lnL AIC AIC wi

ABCDE 40 -8348.9 16777.9 0 0.999

AABCD 48* -8347.7 16791.3 13.4 <0.001

AABCD 32 -8353.9 16794.4 16.5 <0.001

The top three of 104 likelihood-based models tested for growth form and climate occupancy evolution are reported. The best model, based on both AIC and Akaike weights (wi), was a model that assigned a separate rate for the Monocotyledonae (position 1), Magnoliidae (position 2), Superrosidae (position 3), Superasteridae (position 4) and all remaining angiosperms, including the ANA grade, Chloranthales, Ceratophyllales and basal eudicots plus Gunnerales (position 5), respectively. Also listed for each model are the number of parameters, negative log likelihood (2lnL), and DAIC. The asterisk indicates a model where simultaneous changes in any two binary characters were allowed.

CORRIGENDUM doi:10.1038/nature13842 Corrigendum: Three keys to the radiation of angiosperms into freezing environments Amy E. Zanne, David C. Tank, William K. Cornwell, Jonathan M. Eastman, Stephen A. Smith, Richard G. FitzJohn, Daniel J. McGlinn, Brian C. O’Meara, Angela T. Moles, Peter B. Reich, Dana L. Royer, Douglas E. Soltis, Peter F. Stevens, Mark Westoby, Ian J. Wright, Lonnie Aarssen, Robert I. Bertin, Andre Calaminus, Rafae¨l Govaerts, Frank Hemmings, Michelle R. Leishman, Jacek Oleksyn, Pamela S. Soltis, Nathan G. Swenson, Laura Warman & Jeremy M. Beaulieu

Nature 506, 89–92 (2014); doi:10.1038/nature12872 In this Letter, Figs 2 and 3 contained several minor errors, which have now been corrected. In Fig. 2c, we did not include the possible pathway from deciduous and freezing unexposed to evergreen and freezing exposed. This omission slightly alters the relative likelihood of the different pathways out of the evergreen and freezing unexposed state (,2%), but the interpretation is the same. In Fig. 2d, we also note that the arrow leading from large conduits and freezing unexposed to large conduits and freezing exposed and the arrow leading from large conduits and freezing exposed to small conduits and freezing exposed were switched when generating the figure. In general, the scale of the circles (persistence times) and arrows (transition rates) in Figs 2 and 3 were also found to be confusing. We have now corrected Figs 2 and 3 online such that the scale matches a discrete binning of the persistence times and transitions rates for each character state combination. We thank E. Edwards for bringing these issues to our attention. Finally, in Extended Data Table 3, we note an incorrect transition rate was provided for the transition from woody unexposed to woody exposed for the Super- rosidae; the transition rate should be 0.01, not 0.001, and this has also now been corrected online.